Re: Autoreinstallation image is not signed.
On Tue, Jun 3, 2008 at 10:15 PM, John Gilmore [EMAIL PROTECTED] wrote: Specifically, http://dev.laptop.org/ticket/7125. What do people think of the straw man in that ticket? Should we implement it? My comments are in the ticket; let's move the discussion there, where it belongs. Ditto. Briefly: let's fix the real problem (boot fails) before working on an inadequate workaround (deleting user data). --scott -- ( http://cscott.net/ ) ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Autoreinstallation image is not signed.
C. Scott Ananian-3 wrote: http://wiki.laptop.org/go/Olpc-update#USB_upgrade This will not work if the OS is not bootable and no alt-os image is on the disk. C. Scott Ananian-3 wrote: Finally, 8.2 will have better backup/restore functionality, so the real solution then will be reflash+restore. As long as backups are made automagically and often... -FFM -- View this message in context: http://www.nabble.com/Autoreinstallation-image-is-not-signed.-tp17612809p17626313.html Sent from the OLPC Software development mailing list archive at Nabble.com. ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Autoreinstallation image is not signed.
http://wiki.laptop.org/go/Olpc-update#USB_upgrade You can also boot from the ext2 build on an SD card as a recovery mechanism. Finally, 8.2 will have better backup/restore functionality, so the real solution then will be reflash+restore. Please don't use the autoreinstallation key. It has past its use by date. --scott -- ( http://cscott.net/ ) ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Autoreinstallation image is not signed.
On Tue, Jun 3, 2008 at 11:28 AM, ffm [EMAIL PROTECTED] wrote: C. Scott Ananian-3 wrote: http://wiki.laptop.org/go/Olpc-update#USB_upgrade This will not work if the OS is not bootable and no alt-os image is on the disk. We should continue to try very hard not to let the OS become unbootable. If it is unbootable, something Very Wrong should have occurred and there's no guarantee that mount the filesystem and copy off /home will work either. Using a dev key and a rescue disk is probably a much better bet than any attempt at automagic. Please file bugs on ways you've managed to make the OS unbootable, or ways the alt-os image breaks (there are a few); these are likely to get more attention than trying to resuscitate a deprecated tool I wrote for firmware security debugging. That said, there's a separate bug in trac about X not starting when the NAND flash is full. I'm not sure if that's what you're referring to as not booting or not, but we should fix that, too. --scott -- ( http://cscott.net/ ) ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Autoreinstallation image is not signed.
On Jun 3, 2008, at 11:53 AM, C. Scott Ananian wrote: On Tue, Jun 3, 2008 at 11:28 AM, ffm [EMAIL PROTECTED] wrote: C. Scott Ananian-3 wrote: http://wiki.laptop.org/go/Olpc-update#USB_upgrade This will not work if the OS is not bootable and no alt-os image is on the disk. We should continue to try very hard not to let the OS become unbootable. If it is unbootable, something Very Wrong should have occurred and there's no guarantee that mount the filesystem and copy off /home will work either. Using a dev key and a rescue disk is probably a much better bet than any attempt at automagic. Please file bugs on ways you've managed to make the OS unbootable, or ways the alt-os image breaks (there are a few); these are likely to get more attention than trying to resuscitate a deprecated tool I wrote for firmware security debugging. I agree completely with Scott. An interesting data point, however, is that over half of the machines sent for repair in Uruguay are fixed simply by reflashing the machine. This may be an artifact of the old build they are using, but it is a disturbing statistic. In recent months, I've only had to reflash to fix problems which happened right after a previous reflash (#6906). That said, there's a separate bug in trac about X not starting when the NAND flash is full. I'm not sure if that's what you're referring to as not booting or not, but we should fix that, too. --scott -- ( http://cscott.net/ ) ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Autoreinstallation image is not signed.
John, We experienced quite a large number of 'software broken' laptops when we first starting shipping both in Uruguay and in the G1G1 program. I thought one of the things Ivan did in Uruguay was to help them reflash their laptops when they couldn't boot due to journal corruption or other software related reasons. Not sure if this is the same problem. How many do you think are recoverable with software reflash? Have they been recovering these, or have these fallen into one of their other piles of 'broken' laptops? Thanks, Kim On Tue, Jun 3, 2008 at 6:47 PM, John Watlington [EMAIL PROTECTED] wrote: On Jun 3, 2008, at 11:53 AM, C. Scott Ananian wrote: On Tue, Jun 3, 2008 at 11:28 AM, ffm [EMAIL PROTECTED] wrote: C. Scott Ananian-3 wrote: http://wiki.laptop.org/go/Olpc-update#USB_upgrade This will not work if the OS is not bootable and no alt-os image is on the disk. We should continue to try very hard not to let the OS become unbootable. If it is unbootable, something Very Wrong should have occurred and there's no guarantee that mount the filesystem and copy off /home will work either. Using a dev key and a rescue disk is probably a much better bet than any attempt at automagic. Please file bugs on ways you've managed to make the OS unbootable, or ways the alt-os image breaks (there are a few); these are likely to get more attention than trying to resuscitate a deprecated tool I wrote for firmware security debugging. I agree completely with Scott. An interesting data point, however, is that over half of the machines sent for repair in Uruguay are fixed simply by reflashing the machine. This may be an artifact of the old build they are using, but it is a disturbing statistic. In recent months, I've only had to reflash to fix problems which happened right after a previous reflash (#6906). That said, there's a separate bug in trac about X not starting when the NAND flash is full. I'm not sure if that's what you're referring to as not booting or not, but we should fix that, too. --scott -- ( http://cscott.net/ ) ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Autoreinstallation image is not signed.
Hi, That said, there's a separate bug in trac about X not starting when the NAND flash is full. I'm not sure if that's what you're referring to as not booting or not, but we should fix that, too. Specifically, http://dev.laptop.org/ticket/7125. What do people think of the straw man in that ticket? Should we implement it? - Chris. -- Chris Ball [EMAIL PROTECTED] ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Autoreinstallation image is not signed.
chris wrote: Hi, That said, there's a separate bug in trac about X not starting when the NAND flash is full. I'm not sure if that's what you're referring to as not booting or not, but we should fix that, too. Specifically, http://dev.laptop.org/ticket/7125. What do people think of the straw man in that ticket? Should we implement it? so others don't have to look: Here's a straw-man: if disk is full at boot, delete the single largest journal entry, iterate until disk is not full anymore. as a user, i might prefer delete the oldest journal entry, iterate but i'm not convinced either way. paul =- paul fox, [EMAIL PROTECTED] ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Autoreinstallation image is not signed.
Chris, That said, there's a separate bug in trac about X not starting when the NAND flash is full. I'm not sure if that's what you're referring to as not booting or not, but we should fix that, too. Specifically, http://dev.laptop.org/ticket/7125. What do people think of the straw man in that ticket? Should we implement it? Straw man from ticket We're probably going to see this a lot in the field. It might be worth having some recovery logic. Here's a straw-man: if disk is full at boot, delete the single largest journal entry, iterate until disk is not full anymore. end Is there anything that can be thrown away before we start scragging the user's work? Browser caches, or similar things? How much space is needed for a successful boot anyways? Maybe there ought to be a dummy file stored just for the purpose of being thrown away in an emergency. Or throw away least recently used non-core activities, which hopefully could easily be reloaded from the web or a teacher's USB stick. I'd think that throwing away the child's work would be one of the last things we'd want to do. Bob ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Autoreinstallation image is not signed.
On Tue, 3 Jun 2008, Robert Myers wrote: Chris, That said, there's a separate bug in trac about X not starting when the NAND flash is full. I'm not sure if that's what you're referring to as not booting or not, but we should fix that, too. Specifically, http://dev.laptop.org/ticket/7125. What do people think of the straw man in that ticket? Should we implement it? Straw man from ticket We're probably going to see this a lot in the field. It might be worth having some recovery logic. Here's a straw-man: if disk is full at boot, delete the single largest journal entry, iterate until disk is not full anymore. end Is there anything that can be thrown away before we start scragging the user's work? Browser caches, or similar things? How much space is needed for a successful boot anyways? Maybe there ought to be a dummy file stored just for the purpose of being thrown away in an emergency. Or throw away least recently used non-core activities, which hopefully could easily be reloaded from the web or a teacher's USB stick. I'd think that throwing away the child's work would be one of the last things we'd want to do. especially the largest piece of work. there are journal entries that don't store any useful info other then 'this app was used'. those should all be thrown away before any user-generated content is trashed. David Lang ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Autoreinstallation image is not signed.
C. Scott Ananian-3 wrote: We should continue to try very hard not to let the OS become unbootable. If it is unbootable, something Very Wrong should have occurred and there's no guarantee that mount the filesystem and copy off /home will work either. Using a dev key and a rescue disk is probably a much better bet than any attempt at automagic. True, but mount the filesystem and copy off /home is better than nothing. We have to accept that there are builds in the field that have _known_ issues, such as 650. When they occur, (and since users don't update/backup until too late...) we _need_ to have a better solution than you didn't update, you're on your own. -FFM -- View this message in context: http://www.nabble.com/Autoreinstallation-image-is-not-signed.-tp17612809p17636472.html Sent from the OLPC Software development mailing list archive at Nabble.com. ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Autoreinstallation image is not signed.
Specifically, http://dev.laptop.org/ticket/7125. What do people think of the straw man in that ticket? Should we implement it? My comments are in the ticket; let's move the discussion there, where it belongs. John ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Autoreinstallation image is not signed.
On Jun 4, 2008, at 12:56 AM, Kim Quirk wrote: thought one of the things Ivan did in Uruguay was to help them reflash their laptops when they couldn't boot due to journal corruption I gave them a patch that they were able to push out to the machines to restore them to working order _without_ reflashing. Had they had to reflash, they would have lost all the kids' data. -- Ivan Krstić [EMAIL PROTECTED] | http://radian.org ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel